Snap autorefresh issues

Related to Snap auto refresh kills cluster
a few days ago (18th of September), following another snap/lxd auto refresh, some lxc containers crashed again.
Since this happened already 2 times in a few weeks I have a few questions now.
On ubuntu server 16.04, through snap, I unfortunately installed version 3.5, that’s not the LTS version.
How can I get the LTS 3.0.x ? Is there any way to downgrade?
Would you say the LTS version is much more stable?

If I wanted to install lxd 3.0 LTS from the backport repository, is there a procedure to migrate the configuration from snap without losing containers?

I know this touches a few different topics, but they seems all related to me.
Bottom line is that I’m looking for a solution that’s more reliable for production use.

LXD 3.0.x doesn’t get new features so should in general be more stable and a better fit for production environments (so long as you don’t need any of the new features).

Whether you’re using snap or deb doesn’t matter all that much, other than with snaps we can push fixes within hours whereas the deb can take weeks for an update to be pushed through.

LXD doesn’t support downgrading, so whether you want to switch to the 3.0 snap or the 3.0 deb, you will not be able to just switch the LXD version and have it work.

Your best bet, if doable, is to setup a second machine, running LXD 3.0, then use lxc copy or lxc move to move your containers between the two machines, once all moved, you can then reset the source machine whatever way you want before moving the containers back.